Live data from Hacker News

Predicting OpenAI's ad strategy

ossa-ma.github.io

291–300 of 552 posts

Re: Predicting OpenAI's ad strategy

#291
post #268

Earlier quoted context omitted.

>If everyone pays higher wages, there's a greater supply of money for buying stuff / solving problems (assuming the higher wages aren't eaten by rents). No individual form recoups all of the higher wages they pay their workers, obviously, but there's a larger market for the goods of everyone has more money. Does this actually work? Suppose you're on an island where the economy only produces coconuts. How does giving…

This is how we had a major boom in middle-class wealth int he US post WW2. If you are only selling coconuts, a single raw material, yes, you will run into supply constraints such that prices go up. But that isn't how economics works. Your zero-sum economics example is only applicable in short-term scenarios: over the longer term, new industries develop to solve persistent problems that people are willing to pay to so…

>This is how we had a major boom in middle-class wealth int he US post WW2.

The fact that Europe got bombed no doubt helped too, same with the elites being concerned that communism was on the rise and giving workers a better deal in an effort to stave that off.

>Your zero-sum economics example is only applicable in short-term scenarios: over the longer term, new industries develop to solve persistent problems that people are willing to pay to solve.

>Money solves the problems of the people that have the problems. If the problem is 'we need to eat', producers will diversify into new food sources to meet the demand, solving the problem, and capturing the money of the people who have that problem.

I'm not how you got the impression that I thought the economy had to be zero sum. I even specifically mentioned the possibility of more stuff to go around if productivity goes up. That's the problem with your "new industries develop" argument. Unless productivity goes up too, there will only be different stuff, not more stuff overall.

>There is an enormous space of problems people have which cannot be solved due to lack of access to money. Increasing costs in childcare, elder care, and education are good examples.

All of those are service industries that are resistant to scaling, and as a result productivity growth have been abysmal. Giving people more money to spend on those things just means productive capacity is removed from the economy elsewhere. Going back to the coconut economy example, it would certainly be nice if workers could have a maid to do the cleaning or a chef to do the cooking, but you still need people do the cleaning or cooking. At the end of the day you're just shuffling people around, not growing more coconuts.

Re: Predicting OpenAI's ad strategy

#292
post #268

Earlier quoted context omitted.

>If everyone pays higher wages, there's a greater supply of money for buying stuff / solving problems (assuming the higher wages aren't eaten by rents). No individual form recoups all of the higher wages they pay their workers, obviously, but there's a larger market for the goods of everyone has more money. Does this actually work? Suppose you're on an island where the economy only produces coconuts. How does giving…

> Suppose you're on an island where the economy only produces coconuts. This is why nobody takes economists seriously. What you lose in simplifying down to this model is literally everything. The coconut economy has zero predictive power. In the real world, distribution effects dramatically affect the functioning of the economy, because workers are also consumers and owners of capital are siphoning off the purchasing…

>This is why nobody takes economists seriously. What you lose in simplifying down to this model is literally everything. The coconut economy has zero predictive power.

A simplified model is needed otherwise rigorous analysis becomes impossible, and people make handwavy arguments about how paying workers more means they can spend more, which means factories, and it's a perpetual growth machine!

>we’re already massively overproducing just about everything

No we're not. If we weren't, we shouldn't have seen the massive inflation near the end of covid. The supply disruptions hit almost immediately, but it wasn't until the stimmy checks hit that inflation went up.

>our problem is both our wealth and production allocations are borderline suicidal.

If you read my previous comments more carefully, you'd note that I'm not arguing against better wages for workers as a whole, only that contrary to what some people claim, they don't pay for themselves.

Re: Predicting OpenAI's ad strategy

#293

There is something to be said about the state of advertising. Increasingly it seems you must go to the almighty Google or Meta in order to launch any business. We're looking to expand into a new business line and have out grown our pharmacy capacity. The new business line will cost about $2M in software dev, and $3M for the new facility. The advertising budget? $40,000,000 (annual). We can build 10 robotic pharmacies…

> The new business line will cost about $2M in software dev, and $3M for the new facility. The advertising budget? $40,000,000 (annual). The reason the advertising budget is such a high number and a recurring charge is that effective advertising returns an ROI on each dollar spent. If the software budget was increased 10X to $20 million, would the company get 10X as many customers? No. What about the facility? If you…

I had this exact epiphany when I ran a side business selling niche widgets. Every dollar I spent on Facebook ads returned me $5 in revenue (at 40% margin) and it wasn't obvious how much I could spend before that stopped being true.

Re: Predicting OpenAI's ad strategy

#294

Earlier quoted context omitted.

Advertising is a tax that goes to an oligopolistic cartel.

I've never gotten $2 for $1 of tax I paid. At best I'm getting $1 of services back, usually much less.

What “service” do you think you get from the ads you pay for?

Re: Predicting OpenAI's ad strategy

#295
post #241
post #209

Earlier quoted context omitted.

Advertising spend being too high is a symptom of a supply glut. Too many products in the marketplace, not enough consumers to buy them. In a different world where there are higher wages, more people would have more spending power. Then companies wouldn't have to spend as many dollars on advertising, which they could split between higher wages, higher margins and lower prices. Alas, the short-term single-firm directio…

>Maybe Henry Ford was on to something when he shocked the world by paying his employees enough to afford the product they were making (more than doubling many workers' wages)... That's a nice story to tell, but the economics never works out if you do the math. Whatever extra wages you pay, you only get a fraction of that back in increased sales. How much percent of a worker's income do you think is spent on a car? As…

thats an overly simplistic way to look at it. Of course you can never get more money from your employees' purchases than you give them, that makes no sense. The point is using your market power as a large employer to raise market salaries. People will not want to work for your competitors or other sectors if they pay half what you do. So when you rise, naturally other salaries will rise too. And those other workers will also be buying cars.

Whether that makes sense economically is a difficult problem to quantify, especially over any fixed timeframe. I would guess not, at least if your brand isnt insanely strong (on the level of half or more people with enough money would buy it)

Re: Predicting OpenAI's ad strategy

#296

Ads aren't a long-term viable model for tools. Each year it gets more feasible to self host tools (email being the od exception, but there are still many ad free alternatives). Ads shifting vehicles to AI will extend the lifetime a bit, but even still local models are getting better and that's without even much architectural advancement. I don't see an end to advertising all together though, public spaces and enterta…

I'm not so sure. Much like search engines, you can run one yourself or pay Kagi but most people prefer to keep their money and deal with the ads. Streaming services have demonstrated that people have a pretty high tolerance for ads.

Search so far has not been overly pushy with ads. It's easy enough to gain the instinct of scrolling down after each search. There's little incentive for people to seek out an ad free alternative.

That changes with local AI though. There is now incentive to integrate and further develop self hosted search. You can see it happening on AI services already, using their own internal search engines for better reasoning and more accurate results.

I suspect Google's censorship and intentional worsening of search results to increase traffic would've been enough on its own to eventually drive people to self hosted search as it became trivial to setup.

Streaming entertainment is different, there's usually no legal alternatives. Either you pay extra for no ads, or you put up with the ads. You could easily say that streaming services have demonstrated that people don't have a high tolerance for ads as well. One of the major drivers to streaming from cable TV was the lack of ads at the time.

Re: Predicting OpenAI's ad strategy

#297
post #241
post #209

Earlier quoted context omitted.

Advertising spend being too high is a symptom of a supply glut. Too many products in the marketplace, not enough consumers to buy them. In a different world where there are higher wages, more people would have more spending power. Then companies wouldn't have to spend as many dollars on advertising, which they could split between higher wages, higher margins and lower prices. Alas, the short-term single-firm directio…

>Maybe Henry Ford was on to something when he shocked the world by paying his employees enough to afford the product they were making (more than doubling many workers' wages)... That's a nice story to tell, but the economics never works out if you do the math. Whatever extra wages you pay, you only get a fraction of that back in increased sales. How much percent of a worker's income do you think is spent on a car? As…

Your logic is flawed. 8.8% of total wages vs 0% of some lesser number isn’t 8.8% of the difference.

If hypothetically you are paying 99x a year and bumping that to 100x means someone buys an 8.8x product that could be a win.

Re: Predicting OpenAI's ad strategy

#299
post #297
post #241

Earlier quoted context omitted.

>Maybe Henry Ford was on to something when he shocked the world by paying his employees enough to afford the product they were making (more than doubling many workers' wages)... That's a nice story to tell, but the economics never works out if you do the math. Whatever extra wages you pay, you only get a fraction of that back in increased sales. How much percent of a worker's income do you think is spent on a car? As…

Your logic is flawed. 8.8% of total wages vs 0% of some lesser number isn’t 8.8% of the difference. If hypothetically you are paying 99x a year and bumping that to 100x means someone buys an 8.8x product that could be a win.

>If hypothetically you are paying 99x a year and bumping that to 100x means someone buys an 8.8x product that could be a win.

What type of company would this work for? Cars are so ingrained in our society that it's doubtful the marginal Ford (or Tesla) factory worker is going to be buying a $50k car just because they paid better wages.

Re: Predicting OpenAI's ad strategy

#300
post #292

Earlier quoted context omitted.

> Suppose you're on an island where the economy only produces coconuts. This is why nobody takes economists seriously. What you lose in simplifying down to this model is literally everything. The coconut economy has zero predictive power. In the real world, distribution effects dramatically affect the functioning of the economy, because workers are also consumers and owners of capital are siphoning off the purchasing…

>This is why nobody takes economists seriously. What you lose in simplifying down to this model is literally everything. The coconut economy has zero predictive power. A simplified model is needed otherwise rigorous analysis becomes impossible, and people make handwavy arguments about how paying workers more means they can spend more, which means factories, and it's a perpetual growth machine! >we’re already massivel…

> A simplified model is needed otherwise rigorous analysis becomes impossible

If your tools aren’t capable of rigorous analysis of a model that retains enough detail to capture the salient features of the thing they’re trying to model, they’re not the tools for the job.

Post reply on HN